Bluff to peace: How economic dependence promotes peace despite increasing deception and uncertainty

Abstract

Trade–conflict studies have shown that economic dependence can promote peace by costly signaling resolve. However, with higher economic integration, targets also become more vulnerable to coercion and potential challengers are incentivized to bluff. In return, target states may resist more, raising the question of whether trade still promotes peace. I theorize that bluffing does not stoke conflict in this context because the bargaining environment allows states to inform and coerce simultaneously: the factor that renders a threat less credible also restrains states from further escalation. I test this theory’s implications with a structural estimation method and find supporting results.

Keywords

Costly signaling structural estimation trade and conflict trade network

Costly signaling has been a popular explanation of war and peace in recent decades (Fearon, 1994; Schultz, 2001a; Slantchev, 2011). In times of crises, resolute states seek to make their threats credible while the irresolute aim to cheat. To distinguish their threats, resolute states impose substantial costs on themselves, thereby rendering the strategy of mimicking their moves unprofitable for irresolute counterparts. Factors that deter bluffing (i.e. backing down on one’s threat), such as public announcements (Fearon, 1997) and force mobilizations (Slantchev, 2005), can facilitate credible communication and hence promote peace. The commercial peace literature adopts such a costly signaling perspective, adding to the list of factors by suggesting that economic interdependence can promote peace because the willingness to endure economic disruption can credibly signal resolve in the first place (Gartzke et al., 2001; Morrow, 1999).¹

However, in the case of economic dependence, the costly signaling explanation remains incomplete. Instead of solely hurting the challenger, economic coercion hurts the target as well. As such, a costly signal also rearranges the target’s commitment because the target becomes more likely to acquiesce (Morrow, 1999). This in turn generates strategic incentives for some irresolute types to bluff (cf. Gartzke and Li, 2003). Given that bluffing is typically argued to increase the likelihood of war (Fearon, 1995; Ramsay, 2008; Wolford, 2014), it is unclear why increasing economic costs would not incentivize deception, encourage conflict escalation, and ultimately increase the odds of war.

Explaining whether and why this incentive to coerce and deceive stokes or reduces conflict is critical to the further development of commercial liberalism. Above all, this question speaks to the overarching theme of trade’s pacifying effect and the ongoing debate between the opportunity costs and costly signaling theories in the field (Gartzke et al., 2001; Dafoe and Kelsey, 2014). Building on a crisis bargaining model, I argue that economic dependence can still promote peace despite producing a higher likelihood of deception and uncertainty. This is because the bargaining environment of economic dependence allows states to inform and coerce simultaneously.² Existing theories have pointed out one aspect of this simultaneous effect: self-inflicted costs improve credibility and hence generate coercive pressure. My model addresses the other aspect: when states impose costs on an opponent, the coercive effect parallels the informational one. Irresolute states are attracted to initiate a challenge exactly because they anticipate that the target is more likely to concede to avoid costs. As such, when the target stands to suffer more, it will infer that the challenger is less resolved, all else being equal. Yet despite this information, the target is more inclined to concede because it stands to suffer more irrespective of the challenger’s lack of resolve.³ In this scenario, the probability of escalation and war is invariably reduced owing to this increased likelihood of concession. Without this parallel coercive effect (i.e. if trade’s pacifying effect is purely informational), the impact on conflict would be reversed.

My argument that economic dependence allows states to inform and coerce simultaneously can help unify the two primary theories in commercial liberalism. While opportunity costs emphasize the coercive nature of economic dependence, signaling highlights its informational function. I demonstrate that commercial liberalism works via the coercive and informational channels simultaneously. Therefore, while it is important to show how each mechanism operates, it may be misguided to debate which theory dominates.

In addition, this paper speaks to a broader literature of inter-state crisis bargaining. While the mainstream wisdom indicates that increasing uncertainty stokes conflict (Reed, 2003b; Slantchev, 2004), some studies have contended that uncertainty also rewards prudence (Bas and Schub, 2016) and may result in a more peaceful outcome depending on the source of uncertainty (Fey and Ramsay, 2011), whether there is ongoing concern over commitment problems (Wolford et al., 2011; Bas and Schub, 2017), and prior optimistic/pessimistic beliefs (Arena and Wolford, 2012). My study adds to the latter caution and shows that the mechanism of economic dependence can promote peace even when it renders states less certain of an opponent’s resolve.

This paper proceeds as follows. I first discuss the transplantation of the costly signaling theory to the commercial peace literature. I then present a model which demonstrates how the bargaining environment of economic dependence differs from the conventional models, leading to divergent results. These implications are then tested by a structural estimation method that allows belief updating. The final section concludes with implications and limitations.

Costly signaling and commercial peace

Theories of commercial liberalism generally agree on the pacifying effect of economic interdependence (Gartzke et al., 2001; Hegre et al., 2010; Oneal and Russett, 1997; cf. Barbieri, 1996). Two primary explanations have been proposed. The opportunity costs theory focuses on the coercive aspect of economic dependence and reasons that ex post economic costs generate ex ante incentives to avoid conflict. Trade losses owing to conflict add to the costs of war, thereby reducing hostility between states while promoting cooperation (Polachek, 1980; Polachek and Xiang, 2010). Trade can be a vehicle for communication and creates a security community where shared identity suppresses the use of force (Oneal and Russett, 1997, 1999). Lastly, trade can become a cheaper substitute for conquest (Rosecrance, 1986; Gartzke, 2007; Brooks, 2007). In short, economic dependence’s effect is coercive: it constrains belligerents by rearranging their ex ante incentives.⁴

The costly signaling theory, however, contends that if economic disruption only invokes opportunity costs, then its impact on conflict should be indeterminate (Morrow, 1999). That is, although the challenger suffers economically, the target also bears proportional costs and consequently will be more likely to acquiesce. Morrow argues that if trade does promote peace, it should operate by allowing states to communicate their unobservable resolve. With a broader range of costly signals, states can communicate more peacefully and credibly (Morrow, 2003). Gartzke et al. (2001) further develop this costly signaling rationale by arguing that mutual economic dependence help states tell different opponents apart: states that are willing to suffer the costs of disruption are resolved and hence distinguishable from their unresolved counterparts.

Here the reasoning resonates with the wisdom of conventional costly signaling theories. These theories propose several signaling mechanisms, including tying hands which may involve domestic and international audience costs (Fearon, 1994; Sartori, 2002; Schultz, 1998), sinking costs such as troop mobilization (Fearon, 1997; Slantchev, 2005, 2011), and generating autonomous risks that are beyond one’s control (Powell, 1990; Slantchev, 2011). These models agree that a more costly or risky threat indicates the challenger is more resolved. The costly signaling theory in commercial liberalism shares this insight in highlighting that the willingness to bear economic costs helps reveal private information (Morrow, 1999, 2003).

However, the bargaining environment of economic dependence also differs from the conventional models. Specifically, aside from inflicting costs on oneself, it allows challengers to impose costs on an opponent, thereby exploiting the latter’s vulnerability (Barbieri, 1996; Hirschman, 1980; Keohane and Nye, 1977; Peterson, 2014; Wagner, 1988). As such, the incentive to bluff may not be categorically eliminated simply by threatening to cut off trade (Gartzke and Zhang, 2015). Instead, this incentive can grow stronger as the level of dependence rises: as a target becomes more vulnerable, resolute states can step up their demand and irresolute ones may be incentivized to initiate a challenge. In return, the relatively dependent target states should downgrade the credibility of the threat and resist more. Therefore, the same costly signaling mechanism may also contain countervailing forces that stoke conflict and bloodshed.

This incentive to coerce and deceive and its impact on the prospect of peace have not received much attention in trade-conflict studies. However, it is critical to the further development of the literature not least because it can undermine the overarching theme of trade’s pacifying effect. In addition, recent studies also emphasize that imposing costs is a more common practice than inflicting costs on oneself, thereby casting doubt on the underlying logic of costly signaling (Dafoe and Kelsey, 2014). Lastly, it also addresses one area that the opportunity costs theory has overlooked—the issue of incomplete information, which formal theories have identified as a key cause of conflict (Fearon, 1995; cf. Powell, 2006). In other words, the emphasis on vulnerability can be more convincing if it also confronts and explains how the effects of coercion interact with information asymmetry.⁵

In the following section, I present a crisis bargaining model to study how economic dependence affects the choice of bluffing and consequently the likelihood of conflict. I find that heavier economic dependence of the target increases both the likelihood of deception and that of concession. This generates two countervailing forces: (a) the risk of further conflict escalation is reduced because the target become less likely to resist; and (b) the odds of conflict initiation rise because it becomes more lucrative to exploit others’ vulnerability. I then discuss how this mechanism affects the odds of bloodshed and its connections with existing theories.

The model

Two states, 1 and 2, compete over a disputed good owned by State 2. One can interpret this as State 1, the challenger, seeking a policy or territory concession from State 2. State 1 moves first by choosing whether to challenge State 2 (denoted as C or $\tilde{C}$ ), who then moves by choosing whether to escalate the conflict (denoted as E) or not (denoted as $\tilde{E}$ ). If she⁶ concedes, the game ends. If she escalates, State 1 then chooses whether to follow through his initial threat by fighting (denoted as F) or back down from his earlier commitment (denoted as $\tilde{F}$ ). The structure of the game is shown in Figure 1.⁷

Figure 1.

A crisis bargaining model with economic interdependence.

Before I specify the payoffs, a number of notations are in order. Let $v_{i}$ , where $i \in (1, 2)$ , denote each side’s valuation of the disputed good. Write the likelihood of State 1 winning the good by force as p. Denote the military and economic costs of conflict as $c_{i}$ and $b_{i}$ respectively. Finally, let $r_{i}$ represent the audience costs, which State 1 will suffer if he backpedals and State 2 will suffer if she immediately caves.⁸

In the status quo, no state gains or loses anything. That is, each side gains 0. If State 2 concedes immediately, then she loses the good and suffers the audience costs ( $- v_{2} - r_{2}$ ) while State 1 wins the good for free. If State 2 escalates the conflict, then the economic costs are invoked for both sides. If State 1 follows through his threat with force, then the good is allocated by the costly lottery of war where both sides pay the additional military costs as well. That is, the payoffs for fighting can be written as $(p v_{1} - c_{1} - b_{1}, - p v_{2} - c_{2} - b_{2})$ . If, however, State 1 backs down, then both sides avoid the fighting costs and only suffer the economic costs. The payoffs for this outcome can be written as $(- b_{1} - r_{1}, - b_{2})$ .⁹

The sources of uncertainty reside in each side’s valuation of the disputed good. State 1 observes his valuation of the good but is uncertain of his opponent’s ( $v_{2}$ ). Analogously, State 2 knows $v_{2}$ but is unsure of $v_{1}$ . Each side knows the cumulative distributive function of the opponent’s valuation, denoted as $F_{1}$ and $F_{2}$ respectively. For instance, $F_{1} (x)$ denotes the probability that State 1 values the disputed good at a value lower than x.

Equilibrium

The solution of the game is that of Perfect Bayesian Equilibrium. For ease of interpretation, write $v_{1}^{*}$ and $v_{2}^{*}$ as the benchmark that States 1 and 2 are resolved. Specifically, if $v_{1} > v_{1}^{*}$ , State 1 prefers war to concession ( $\tilde{F}$ ). If $v_{2} > v_{2}^{*}$ , State 2 prefers war to concession ( $\tilde{E}$ ). Mathematically, $v_{1}^{*} = (c_{1} - r_{1}) / p$ and $v_{2}^{*} = (c_{2} + b_{2} - r_{2}) / (1 - p)$ . In addition, write ${\hat{v}}_{1}$ and ${\hat{v}}_{2}$ as the benchmark for State 1’s and 2’s escalation strategy. That is, if $v_{1} > {\hat{v}}_{1}$ , State 1 provokes; otherwise, he opts to not provoke. If $v_{2} > {\hat{v}}_{2}$ , State 2 further escalates the conflict; otherwise, she concedes. Finally, denote State 2’s belief that State 1 is resolved after observing a challenge as $α$ . By Bayesian rules, $α = (1 - F_{1} (v_{1}^{*})) / (1 - F_{1} ({\hat{v}}_{1}))$ . The equilibrium is as follows.

Proposition 1. Different types of the two states play by the following rule:

State 1 chooses to challenge if his valuation of the disputed good is greater than ${\hat{v}}_{1}$ , and restrains if it is not.

State 2 escalates the conflict if her valuation of the disputed good is greater than ${\hat{v}}_{2}$ , and concedes if it is not. State 2 updates her belief that State 1 is resolved to $α$ when she sees a threat being issued.

The values of ${\hat{v}}_{1}$ , ${\hat{v}}_{2}$ , and $α$ are defined implicitly by the following functions:

{\hat{v}}_{1} = \frac{1 - F_{2} ({\hat{v}}_{2})}{F_{2} ({\hat{v}}_{2})} (b_{1} + r_{1})

{\hat{v}}_{2} = \frac{α c_{2} + b_{2} - r_{2}}{1 - p α}

α = \frac{1 - F_{1} (v_{1}^{*})}{1 - F_{1} ({\hat{v}}_{1})}

Comparative statics

The solution of the proposition’s comparative statics is a bit involved and is therefore left to the Online Appendix. Importantly, we have the following results:

\frac{\partial {\hat{v}}_{1}}{\partial b_{1}} > 0 \frac{\partial {\hat{v}}_{2}}{\partial b_{1}} > 0 \frac{\partial α}{\partial b_{1}} > 0

\frac{\partial {\hat{v}}_{1}}{\partial b_{2}} < 0 \frac{\partial {\hat{v}}_{2}}{\partial b_{2}} > 0 \frac{\partial α}{\partial b_{2}} < 0

Recall ${\hat{v}}_{1}$ denotes the benchmark of State 1 issuing a challenge. A higher (lower) value of ${\hat{v}}_{1}$ means the requirement is more (less) stringent, i.e. fewer (more) types of State 1 find the choice attractive. It, therefore, captures the probability that State 1 is bluffing. Analogously, ${\hat{v}}_{2}$ denotes the likelihood of State 2 conceding to a threat, where a higher (lower) ${\hat{v}}_{2}$ means it takes a higher (lower) valuation for State 2 to resist. The variable $α$ denotes the belief of State 2 that State 1 is resolved after observing his threat.

The above comparative statics have a number of critical implications. First, they confirm the conventional wisdom that self-inflicted costs increase the credibility of threats. They also show that the informational and coercive functions are intertwined: a target is coerced (becomes less likely to resist) because the threat becomes more credible. To see this, observe that when $b_{1}$ increases, both $α$ and ${\hat{v}}_{2}$ increase. A higher value of $α$ means that State 2 believes the threat is more likely coming from a resolved type. A higher value of ${\hat{v}}_{2}$ means that States 2 is less likely to resist the threat.

Second, the results reveal another drastically different mechanism: imposing costs on a target induces concession even when the threat’s credibility drops. To see that, observe that when $b_{2}$ increases, $α$ decreases while ${\hat{v}}_{2}$ increases. The former suggests that State 2 downgrades the credibility of the threat (because more irresolute types, i.e. a lower benchmark of ${\hat{v}}_{1}$ , are attracted to challenge her) while the latter suggests that State 2 is less likely to fight back. In this regard, economic vulnerability rewards the bluffer even when the credibility of his threat drops. To put it differently, higher trade dependence of the target means a relatively higher chance of successful bluffing because target states give in more readily despite their being more suspicious of the challenger. As such, a larger proportion of potential bluffers, who originally viewed the disputed good as not worth fighting for and chose to maintain the status quo, are now willing to initiate a conflict exactly because they anticipate this shift of bargaining advantage. Therefore, I have the following hypotheses.

Hypothesis 1. The more dependent the target state’s economy ( $b_{2}$ ) is on the challenger, the more likely the target will be to concede to a challenge.

Hypothesis 2. The more dependent the target state’s economy ( $b_{2}$ ) is on the challenger, the less likely it is that the target will perceive the threat to be credible.

Given that we should expect more challenges initiated by an irresolute type, I have an additional hypothesis.

Hypothesis 3. The more dependent the target state’s economy ( $b_{2}$ ) is on the challenger, the more likely the challenger will be to back down from his provocation.

Third, even when it becomes more lucrative for State 1 to bluff, the likelihood of war is still reduced. This pacifying impact can be attributed to both the informative and coercive functions of economic dependence. The likelihood of war can be written as

\Pr (War) = F_{1} (v_{1} > {\hat{v}}_{1}) \cdot F_{2} (v_{2} > {\hat{v}}_{2}) \cdot F_{1} (v_{1} > v_{1}^{*} | v_{1} > {\hat{v}}_{1})

which can be simplified as $F_{2} (v_{2} > {\hat{v}}_{2}) \cdot F_{1} (v_{1} > v_{1}^{*})$ . Note that the latter part is a constant. Hence, the likelihood of war can be evaluated directly by $F_{2} (v_{2} > {\hat{v}}_{2})$ . Observe that when either $b_{1}$ or $b_{2}$ increases, ${\hat{v}}_{2}$ increases as well, resulting in a lower $F_{2} (v_{2} > {\hat{v}}_{2})$ . In other words, the possible increase in the probability of war is ultimately overwhelmed by the target’s tendency of concession.

Intuitively, there are two opposing forces affecting the odds of war: (a) the pool of conflict onset expands as more potential challengers are incentivized to initiate a conflict; and (b) the likelihood of escalation decreases (otherwise, bluffers would not be incentivized in the first place).¹⁰ While (b) reduces the odds of war, (a) may increase them. The latter’s impact is restrained because it is inflated by unresolved types, which by definition will back down once their bluff is called. As such, the effects of (a) become inconsequential and the odds of bloodshed are ultimately reduced as the target becomes more likely to concede. Therefore, I have the following hypothesis.

Hypothesis 4. The more dependent the target state’s economy ( $b_{2}$ ) is on the challenger, the less likely it is that the conflict will end up in war.

Importantly, the effects of $b_{1}$ and $b_{2}$ link the model with the existing theories of commercial peace. The variable $b_{1}$ resonates with the theory of costly signaling (Gartzke et al., 2001; Morrow, 1999; Reed, 2003a): State 2 is coerced by an informative threat because self-inflicted costs improve credibility. Analogously, $b_{2}$ echoes the reasoning of opportunity costs (Polachek and Xiang, 2010; Dafoe and Kelsey, 2014): State 2 concedes to a threat that capitalizes on her economic vulnerability (even though it is less credible). Each theory underscores either the informational or coercive function of economic dependence. My model showcases that they should instead be united because the bargaining environment allows both functions to work at the same time: informative threats bear coercive impacts and coercive threats can compensate for the lack of credibility.

To sum up, my model suggests that economic dependence still promotes peace despite a higher likelihood of deception because target states’ economic vulnerability simultaneously encourages deception and concession. Challengers deceive exactly because they anticipate targets to concede more willingly. As such, the odds of war reduce even though the credibility of threat also drops. Granted, no observer can know a priori whether a challenger is bluffing or not. As such, it is extremely difficult to differentiate whether a concession by the target is driven by the credible revelation of resolve or by bluff success. However, if the empirics do indicate that higher levels of economic dependence are correlated with a higher likelihood of concession and deception simultaneously, that will provide substantial support to my theory.

Research design

To test the above hypotheses, I apply the structural estimation method for signaling games (Whang et al., 2013; Whang, 2010), which has the advantage of estimating all four pairs of relationship in the hypotheses simultaneously (i.e. economic dependence’s relationship with concession, deception, belief updating, and the likelihood of war). I choose this approach because of the need to model strategic interaction (Signorino, 1999; Smith, 1999; Sartori, 2003).¹¹ Moreover, conventional empirical approaches to estimate belief updating (H2) would require a monumental effort to sift through how leaders interpret and revise their beliefs throughout each conflict process, which can be further complicated by the fact that historical records tend to bias against bluff success (bluffs uncalled are less likely to be recognized and recorded). Before such data are available, the structural estimation represents at least a compromise for us to gain some traction to estimate H2 systematically.¹²

While details of the structural estimation method can be involved, the underlying rationale is not. Intuitively, it is similar to the maximum likelihood estimation, which searches for the parameters of a probability density function that maximizes our likelihood of observing the sample data. Analogously, the structural estimation requires us to write out and maximize the joint probability function of observing the data. The catch is that we have to incorporate all outcomes of the strategic interaction simultaneously. Specifically, as shown in Figure 1, we observe one of four different outcomes for a crisis: SQ (status quo), CD (concession by the target), BD (backing down by the challenger), and WAR. For each outcome, we specify a set of variables (X) as well as their respective parameters ( $β$ ) that contribute to each player’s payoff ( $β X$ ). These payoffs contribute to the likelihood of observing each outcome by an underlying probability function: $\Pr (y | β, X)$ , where $y \in {SQ, CD, BD, WAR}$ . Using this information, we can write up the joint likelihood function of observing the outcomes. Finally, we use the maximum likelihood estimation method to estimate the underlying probability function and the parameters that contribute to each outcome (for details, see Whang, 2010).

Model specification

The structural estimation method requires that we specify the variables of interest in each node when a player has to make a choice. Since there are four outcomes and two players, there are eight payoff functions to specify. However, State 2’s payoff for remaining at the status quo cannot be estimated and is therefore omitted (Whang, 2010; Whang et al., 2013). Also, her payoff when State 1 backs down from his threat is normalized to 0. As such, I only need to specify six payoff functions. That is, I need to specify which variables affect State 1’s payoffs over SQ, CD, BD, and WAR. I also need to specify which variables affect State 2’s payoff over CD and WAR.

The details of model specification are shown in Table 1. I include each side’s economic dependence in their own decision node, given that the focus of this article is to examine economic dependence’s impacts on a state and its opponent’s strategic choices. That is, State 1’s dependence is included in the functions where he has to choose whether or not to change the status quo, whether or not to back down when the opponent resists his threat. Similarly, State 2’s dependence is included in her choice over concession or conflict escalation.

Table 1.

Model specification. SQ (status quo), CD (concession by the target), BD (backing down by the challenger), and WAR.

Player	Outcome	Variables
1	SQ	Economic Dependence of State 1, Alliance, Contiguity, Major Power Status of State 1, Major Power Status of State 2
1	CD	Power Ratio
1	BD	Economic Dependence of State 1, Democracy of State 1
1	WAR	Power Ratio, Contiguity, Defense Pacts of State 1
2	CD	Economic Dependence of State 2, Democracy of State 2
2	WAR	Power Ratio, Contiguity, Defense Pacts of State 2

I include major power status for both states in the status quo function for State 1 given that major powers tend to interact more with and are less likely to be challenged by other states.¹³ For State 1’s payoffs when 2 concedes, I add power ratio because extracting a concession from a stronger opponent tends to be more lucrative. I use the democracy level to capture audience costs (Fearon, 1994; Schultz, 2001b) when State 2 concedes immediately or when State 1 backs down from his threat. Both states’ payoff functions for war are estimated by power ratio, contiguity, and their number of defense pacts.

The dependent variables

To test the hypotheses with the structural estimation method, I need a dataset that captures all four different outcomes: SQ, CD, BD, and WAR. I use a version of the Militarized Interstate Dispute (MID) data (Gibler et al., 2016), which applies a strict reading of the coding rules to revise the original dataset. This revised dataset drops 251 cases that do not meet MID coding rules, merges 72 MIDs, and makes major changes to 234 disputes and minor changes to 1009 disputes. Given that the model introduced here bears implications for disputes that can potentially disrupt normal economical exchange, I find this version of data appropriate for testing the hypotheses.¹⁴

Using this revised dataset, I code the four outcomes by consulting the originator, outcome, and hostility level variables. I code an outcome as status quo if the hostility level is 1 (i.e. no militarized action), concession when a target (i.e. a state which is not an originator of dispute) yields to a threat, and war when one state’s hostility level reaches 5. When the target does not comply and the initiator does not follow through with war (including stalemate cases), it counts as backing down.¹⁵

The unit of analysis is directed dyad-year. Owing to missingness, the sample size varies as I include different dependent and independent variables. For the main model, there are 1041 observations from 1960 to 2010.¹⁶

The independent variables

The key independent variable is economic dependence, which I operationalize by two steps. First, I follow one conventional operationalization and generate a rough measurement of dependence using dyadic trade divided by a state’s GDP (Oneal and Russett, 1997).¹⁷ Second, I weigh this variable by a state’s vulnerability to economic coercion. There are two aspects to consider here: (a) larger economies tend to be more influential; and (b) all else being equal, states that are more integrated into the global economy have more alternatives and are hence less vulnerable to economic coercion. As such, they tend to value a given trade relationship less (Barbieri, 1996; Crescenzi, 2003a; Dorussen and Ward, 2010; Hirschman, 1980; Peterson, 2011, 2014).

In line with the recent development of network analysis, I use the tnet package in R to generate the closeness measurement as a proxy of states’ economic vulnerability (Kinne, 2012, 2014; Opsahl et al., 2010; Peterson, 2018). I choose to weigh trade networks by trade volume, instead of trade dependence.¹⁸ Not using trade dependence can potentially bias the measurement toward big economies. However, this is a desirable tradeoff as my primary goal of generating the vulnerability measure is to use it to weigh the economic dependence measurement. This way, the weighted variable can strike a balance between the size and integration level of an economy.¹⁹ That said, I also rerun the model based on trade networks weighted by economic dependence. Most models show similar patterns, though in some cases results turn less significant. Using an alternative centrality measure, the eigenvector centrality, I only have comparable results for the trade share measurement.²⁰

To be a desirable proxy for states’ economic vulnerability, the closeness measurement should demonstrate a number of attributes. First, it should be able to capture the influence of international and domestic events. Second, as mentioned previously, bigger and more influential economies typically should be less vulnerable.²¹ Last, as highlighted by Kinne (2012), it should not be biased against small yet highly integrated economies.

Weighting trade networks by trade volume can produce such a measurement. As an illustration, I plot the closeness values and rankings of two pairs of countries in Figure 2.²² The first pair consists of big economies (the US and China), while the second pair consists of relatively small ones (Belgium and Bahamas). First, the closeness measurement appears to capture the influence of both international and domestic events quite well. As is demonstrated in Figure 2a and c, we can see that the general trend for the past century is toward more integration and globalization. In addition, the two world wars have a dramatic influence on the closeness measurement as the countries’ values drop sharply in both periods. In terms of domestic events, take the Chinese economy as an example. Looking at the period after World War II, we can see the Chinese economy taking a tip in the 1960s (the Cultural Revolution) and then steadily improving its status from the 1980s (the Reform and Opening-up led by Deng Xiaoping). Second, the measurement also showcases that bigger economies (under similar levels of integration) are generally more influential and less vulnerable. For instance, the economies of the US and Belgium are both open and well integrated (Figure 2b and d shows that they are both in the top 10 list most of the time). Still, the US economy is consistently regarded as being relatively less vulnerable than Belgium’s.

Figure 2.

Value and ranking of the closeness measure. The x-axis denotes time, while the y-axis denotes the value or ranking of the closeness centrality measure in a given year. In the upper (lower) panels, the yellow solid line denotes the rankings of the US (Belgium) over time, while the gray dashed line denotes that of China (Bahamas).

Finally, the measurement does not bias much against small and highly integrated economies. Figure 2d shows that the rankings of Belgium, bar the two world war periods, have been consistently in the top 10. If we compare the values of Belgium’s closeness measure against those of the US (Figure 2c and a), we can see that they do not differ much. Similar patterns exist for other relatively small and highly integrated economies such as Singapore, Japan, South Korea, etc.²³ In comparison, small countries with a limited level of integration (e.g. the Bahamas in Figure 2d) are becoming comparatively more vulnerable over time.

In this regard, I believe that weighting trade networks by trade volume can capture the afore-mentioned dual aspects of economic vulnerability (economic size and integration). I then proxy the level of economic dependence by taking the product of trade dependence and the exponential of the negative value of closeness (trade dependence * exp(-closeness)). Although a simple operationalization choice, I find it reasonable in that the exponential is bounded²⁴ and it captures the core idea that as a state becomes less vulnerable in its trade network it values a given trade relationship less.

As for the controls, I follow some conventional choices. I use Polity IV (Marshall et al., 2002) to proxy audience costs. I also use COW’s data on alliance (v 4.1, Gibler, 2009), contiguity (Stinnett et al., 2002),²⁵ major power status, and power ratio (v 5.0, Singer et al., 1972) to measure the relative military strength and the costs of war.

Results

The coefficient estimates for the model are shown in Table 2, Model 1. Each row indicates the effect of a variable on a state’s payoff for one corresponding outcome. A positive (negative) value indicates that the variable increases (decreases) the respective payoff. The main row of interest (highlighted) is “CD2:TradeDepend2”, which represents the effect of State 2’s economic dependence on her payoff of concession. The result is positive and highly significant, suggesting that as a target becomes more dependent she finds the choice of concession to be more attractive.

Table 2.

Structural estimation results: SQ (status quo), CD (concession by the target), BD (backing down by the challenger), and WAR. Model 1 uses data as introduced in the paper, Model 2 uses the International Crisis Behavior data, Model 3 uses political affinity to generate SQ, and Model 4 uses trade share without weighting by trade network.

	(1)Main model	(2)ICB	(3)SQ by affinity	(4)Trade share
SQ1: TradeDepend1	−0.34	3.19	0.15	0.15
	(−3.44, 2.75)	(−4.70, 11.08)	(−1.95, 2.24)	(−0.18, 0.48)
SQ1: Alliance	−0.03	0.04	0.01	0.01
	(−0.13, 0.07)	(−0.07, 0.15)	(−0.06, 0.08)	(−0.05, 0.06)
SQ1: Contiguity	−0.55^***	−0.23	−0.56^***	−0.31^***
	(−0.77, –0.32)	(−0.64, 0.17)	(−0.78, –0.34)	(−0.44, –0.18)
SQ1: Major1	−0.003	0.27	0.36^***	−0.10
	(−0.12, 0.12)	(−0.06, 0.60)	(0.16, 0.55)	(−0.23, 0.03)
SQ1: Major2	−0.04	0.14	0.12^*	0.03
	(−0.16, 0.09)	(−0.12, 0.41)	(−0.002, 0.24)	(−0.03, 0.09)
CD1: Constant	−0.74	−0.11	−0.55^**	0.51
	(−5.17, 3.69)	(−3.26, 3.03)	(−0.99, –0.10)	(−0.81, 1.83)
CD1: PowerRatio	0.17	0.18	−0.15	−0.24
	(−3.59, 3.93)	(−2.59, 2.96)	(−1.26, 0.96)	(−1.10, 0.62)
BD1: Constant	0.32^*	−1.03^***	−0.22^*	0.44^***
	(−0.04, 0.69)	(−1.71, –0.34)	(−0.46, 0.03)	(0.25, 0.63)
BD1: TradeDepend1	0.14	−4.62	0.03	−0.54^**
	(−3.83, 4.12)	(−20.06, 10.81)	(−5.02, 5.09)	(−1.04, –0.03)
BD1: Democracy1	0.02^*	0.01^**	−0.03^***	0.01^*
	(−0.001, 0.03)	(0.001, 0.03)	(−0.04, –0.02)	(−0.001, 0.02)
War1: Constant	0.13	−0.25*	−0.07	0.04
	(−0.39, 0.64)	(−0.54, 0.04)	(−0.37, 0.24)	(−0.44, 0.52)
War1: PowerRatio	0.12	0.20	0.04	0.08
	(−0.10, 0.34)	(−0.07, 0.48)	(−0.14, 0.22)	(−0.03, 0.19)
War1: Contiguity	−0.45^*	−0.36^**	−0.48^***	−0.28^**
	(−0.92, 0.03)	(−0.70, –0.02)	(−0.71, –0.25)	(−0.50, –0.06)
War1: Defense1	−0.0004	0.002	0.003	−0.003^**
	(−0.003, 0.002)	(−0.001, 0.01)	(−0.001, 0.01)	(−0.01, –0.0004)
CD2: Constant	−0.06^*	−0.16^***	−0.01^***	−0.12^***
	(−0.12, 0.003)	(−0.17, –0.15)	(−0.01, –0.003)	(−0.13, –0.11)
CD2: Trade Depend2	0.60^***	1.22^***	0.03^***	0.09^**
CD2: Trade Depend2	(0.35, 0.85)	(1.20, 1.23)	(0.01, 0.05)	(0.01, 0.16)
CD2: Democracy	−0.001	0.01^*	0.0003^*	−0.002^***
	(−0.004, 0.002)	(−0.002, 0.03)	(−0.0000, 0.001)	(−0.002, –0.002)
War2: Constant	0.27	1.08	0.42^*	0.21
	(−1.42, 1.97)	(−0.54, 2.69)	(−0.06, 0.90)	(−0.15, 0.56)
War2: PowerRatio	0.67	−0.43	0.42	0.12
	(−0.66, 2.00)	(−2.28, 1.42)	(−0.16, 1.00)	(−0.32, 0.56)
War2: Contiguity	0.18	−0.37	1.38^***	−0.16
	(−0.50, 0.86)	(−1.04, 0.30)	(0.98, 1.78)	(−0.39, 0.06)
War2: Defense2	0.02	0.04^*	−0.02^***	0.01^***
	(−0.01, 0.04)	(−0.004, 0.08)	(−0.03, –0.01)	(0.004, 0.02)
Observations	1041	800	3256	2170
Log-likelihood	−830.64	−599.07	−1496.83	−2048.66

Note: ^*p < 0.1;^**p < 0.05; ^***p < 0.01.

The impact on payoffs is not equal to the probability of outcomes. This is because while it directly affects a state’s choice on one corresponding outcome, it also indirectly affects the other’s strategic calculation. To demonstrate both the direct and indirect impacts of the target’s economic dependence, I plot four pairs of relationship in Figure 3. Specifically, I plot the impacts of the target’s dependence on the probabilities of the challenger being irresolute (deception), the likelihood of the target’s concession, the odds of war, and the degree of belief updating. Note that these are not conditional probabilities and only represent states’ strategic consideration. Also, the graph is plotted by holding the other variables at certain values. Therefore, the following interpretation focuses primarily on the general pattern, instead of the substantial changes of probabilities.

Figure 3.

Impact of the target’s economic dependence. This plot is composed by varying State 2’s dependence from its minimum to the 90th percentile. The two states are held as non-contiguous major powers. Other continuous variables are held at their median values. The y-axis in panels (a) and (b) denotes unconditional probabilities of concession and backing down respectively. The y-axis in panel (c) denotes the differences between posterior and prior probabilities (beliefs): negative (positive) values indicate that the target downgrades (upgrades) her belief on the likelihood of the challenger being resolute. The y-axis in panel (d) denotes conditional probabilities of war.

The direct impact of economic dependence is plotted in Figure 3a. As a target’s economic vulnerability deepens, it becomes more likely to concede. This impact indirectly rearranges the challenger’s incentives. Specifically, more irresolute challengers are attracted to issue a threat, meaning that when their threats are resisted they are more likely to back down. As such, the likelihood of deception rises as the target’s dependence increases, as shown in Figure 3b.

In contrast to these two pairs of positive relationship, trade dependence’s impacts on the credibility of a threat and war are both negative. Figure 3c shows the change of belief. The value of belief updating is composed by deducting the prior belief of the challenger being resolute from the posterior belief. The negative values indicate that the target always downgrades the credibility of a threat that exploits her economic vulnerability. Moreover, as this vulnerability increases, the threat’s credibility drops further. Despite this lack of credibility, the pacifying impact of dependence still persists. As shown in Figure 3d, the odds of war are invariably reduced as the target’s dependence deepens.

The confidence intervals of all four pairs of relationship widen as the level of trade dependence rises. This can be attributed to the nature of the data—I do not have many cases where a state’s trade is heavily dependent upon a specific country. Therefore, the estimation for higher levels of dependence is less certain. However, these pairs of relationship do offer support to the four hypotheses derived earlier. In short, economic dependence decreases the likelihood of war despite a heavier cloud of deception. This is because the bargaining environment allows states to inform and coerce simultaneously. When states are informed that a threat is less credible, they do not resist more under the context of economic dependence because the exact informational mechanism is coupled with a coercive counterpart.

Robustness checks

To make sure that the results are not driven by arbitrary choices of either data or operationalization, I perform a number of robustness checks. I use (a) the International Crisis Behavior data to generate an alternative set of dependent variables, (b) the UN general assembly voting data to generate a different status quo variable, (c) the trade share measurement to proxy economy dependence, (d) the dependence measurement without weighting the trade network and (e) alternative weights and centrality measurement for the trade networks. I show some of the results in Models 2–4 in Table 2.²⁶ The general pattern is confirmed by most results: a target’s economic dependence simultaneously encourages deception and concession. Relatedly, it also promotes peace despite a lower credibility of threats.

Dual functions of economic interdependence

Combined with the existing wisdom in commercial peace literature, the above results suggest that inflicting or enduring economic costs on oneself signals resolve and can convince irresolute adversaries to quit, while imposing costs on an opponent can test the target’s determination and nudges it toward acquiescence despite possible lack of credibility. This does not necessarily indicate that leaders should or will ignore the negative impact on credibility.²⁷ One practical way to complement economic coercion is to increase the publicity. China, for instance, publicly destroyed 35 tons of Philippine bananas in March 2016 in response to the latter’s claim of the South China Sea dispute in the International Court of Arbitration.²⁸ If China were only concerned about the coercive effect, then this publicity is meaningless.

More broadly, when states flex their economic muscles, the strategic calculations are not solely about either coercion or signal. First, imposing and enduring economic costs are two sides of the same coin. That is, when a challenger seeks to coerce, its target can endure the costs to signal resolve. For instance, when South Korea agreed to install the Terminal High Altitude Area Defense system in 2017, China rallied nationwide support to divert its tourists and boycott South Korea’s stores and products. By some estimates, Chinese sanctions cost South Korea around 0.5% of its GDP, much more than it cost Beijing (around 0.02%).²⁹ To be sure, the coercive effects were substantial: South Korea companies and citizens eagerly urged the government to end the spat.³⁰ However, Seoul chose to endure the economic and political pressure. This in turn convinced China that South Korea was resolute on the issue and prodded Beijing to blink later that year.³¹

Second, states typically evaluate the informational and coercive impact concurrently. Consider Britain’s reaction toward US coercion during the Suez Crisis. If the impact of denying London’s access to the International Monetary Fund were purely coercive, then Britain should not have retreated, at least not immediately. Indeed, Britain’s capacity and willingness to endure the economic disruption was genuine: when Macmillan was informed on the threats of the balance of payments, he convinced himself that Britain was ‘pretty well armed for Suez’. In late October, the prime minister told his colleagues that he expected to lose $300 million and his government’s policy was to see things through (Turner, 2014, p. 119). On the eve of British retreat, there was no immediate need for Britain to draw the Fund. In fact, pressure on sterling had eased, which might have been further improved if the Canal were captured (Fforde, 1992).

Although this does not suggest the coercive effect was immaterial, it does showcase that it was not the only factor in play. In particular, the recognition of the true intention of the US played an important role. Prior to the crisis, British leaders mistakenly believed that they would have US support (without which they also firmly believed they would end the military course). Even after Eisenhower’s clear correspondence and the deployment of the Sixth Fleet, they still retained the belief that the US would not oppose. At worst, the US would ‘lament publicly and do nothing’ (Steed, 2016, p. 67). The misinformation was further amplified by Downing Street’s inclination to interpret ‘what they wanted to hear’ from their American counterparts’ statements (McCourt, 2014, p. 70). US warnings were read as a possible acceptance of a fait accompli, if delivered speedily. Although denying International Monetary Fund access did not bear an immediate coercive impact, the willingness of the US to publicly threaten the economic exchange with a critical ally updated the prime minister’s prior belief and convinced him that Amercian goodwill ‘could not be obtained’ without an immediate cease-fire and retreat (Turner, 2014, p. 123).³²

Conclusion

I have argued that the bargaining environment of economic interdependence allows states to inform and coerce simultaneously. This is important because the field has been interpreting the two as opposing mechanisms: states can either inform or coerce, but not both. Focusing on target states’ vulnerability, I argue that neither mechanism can dominate. Specifically, if commercial peace works solely via the signaling channel, then a higher level of economic dependence can indicate a less credible threat, resulting in more conflict escalation and bloodshed. I argue that this is not the case in the context of economic dependence because a coercive channel parallels the informational one. The exact factor that indicates a lower credibility also constrains, leading to a lower likelihood of escalation and bloodshed. Analogously, if economic dependence only coerces, then imposing economic losses on oneself makes little sense as it will only drain away one’s bargaining leverage. Therefore, instead of debating the merits of either theory, we should interpret the opportunity costs and costly signaling theories as two parallel mechanisms.

My theory also reveals the nuances of the informational mechanism: credibility itself does not dictate states’ decision over conflict escalation. It also depends on the sources that generate the uncertainty and propel the belief updating process. In the context of economic dependence, self-inflicted costs increase the credibility of threats, resulting in a lower likelihood of conflict escalation. In this scenario, higher (lower) credibility is associated with less (more) conflict. In contrast, the association between credibility and conflict is reversed when it comes to imposed costs: imposing economic pains on a target forces it to concede, despite a lower credibility.

That said, additional research is still needed. To begin with, the structural estimation method is an indirect way of testing the information updating process. While it reveals how states should rationally revise their belief, it is still a step away from what they do in reality. Further research, such as an experimental design, can provide a closer look at the informational mechanism. Second, given that states and leaders typically seek to impose asymmetrically higher costs on an adversary while reducing or avoiding costs on themselves (Dafoe and Kelsey, 2014; Gartzke and Westerwinter, 2016), an important question in order is how does imposing asymmetric costs affect the prospect of peace?³³ That is, there are two opposing forces: (a) reducing one’s economic pains reveals lack of resolve, incentivizes resistance, and stokes conflict; and (b) imposing higher economic costs on a target induces concession and suppresses conflict. Exactly how the two forces interact is beyond the scope of this paper. However, if we consider both the coercive and informational functions of economic ties, the effects of asymmetric dependence on conflict are not monotonic and deserve further investigation (Crescenzi, 2003b).

Third, other factors such as time and reputation have been abstracted away in this paper. However, it is critical to study how they interact and modify states’ strategic calculations. For instance, when North Korea stepped up its nuclear ambitions in late 2017, China imposed substantial sanctions against the regime.³⁴ In response, Kim Jong-un effectively dialed back his hostile stance against the US and South Korea. Although no observers can be sure of the reasonings in Kim’s mind, my theory suggests he would question Chinese leaders’ resolve. Although he bowed to the immediate economic pressure, he would likely continue his nuclear ambitions while expecting China to ease the pinch over time. The crisis bargaining model in this paper focuses on the short-term immediate effect of coercion and cannot address the trade-off between short-term concessions and long-term benefits. Relatedly, reputational effects can kick in as we consider the impact of time: leaders may intentionally choose a suboptimal strategy in order not to invite future coercion from other states (Crescenzi, 2007; Peterson, 2013). That said, this paper suggests that when examining these further extensions (asymmetry, time, and reputation) we should consider the coercive and signaling mechanisms jointly.

Supplemental Material

Bluff_Econ_OnlineAppendix – Supplemental material for Bluff to peace: How economic dependence promotes peace despite increasing deception and uncertainty

Supplemental material, Bluff_Econ_OnlineAppendix for Bluff to peace: How economic dependence promotes peace despite increasing deception and uncertainty by Yuleng Zeng in Conflict Management and Peace Science

Supplemental Material

DA-RT – Supplemental material for Bluff to peace: How economic dependence promotes peace despite increasing deception and uncertainty

Supplemental material, DA-RT for Bluff to peace: How economic dependence promotes peace despite increasing deception and uncertainty by Yuleng Zeng in Conflict Management and Peace Science

Footnotes

Acknowledgements

The author wishes to thank Taehee Whang, Elena V. McLean, and Douglas W. Kuberski for kindly sharing their code, Timothy Peterson, Tobias Heinrich, Brad Epperly, Yu-hsien Sun, Mathew Lawson, Kaitlin Mcclamrock, Casey Crisman-Cox, Michael Murphree, attendees at the 2019 SPSA conference, and the editors and five anonymous reviewers from CMPS for reading over the manuscript and offering insightful suggestions. I am also thankful for the financial support from the SPARC Graduate Research Grant from the Office of the Vice President for Research at the University of South Carolina. All errors remain mine.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research received financial support from the SPARC Graduate Research Grant from the Office of the Vice President for Research at the University of South Carolina.

Supplemental material

All data, replication materials, and instructions regarding analytical materials upon which published claims rely are available online through the SAGE CMPS website: .

ORCID iD

Yuleng Zeng

Notes

References

Anderton

Carter

(2001) The impact of war on trade: An interrupted times-series study. Journal of Peace Research 38(4): 445–457.

Arena

Wolford

(2012) Arms, intelligence, and war. International Studies Quarterly 56(2): 351–365.

Barbieri

(1996) Economic interdependence: A path to peace or a source of interstate conflict? Journal of Peace Research 33(1): 29–49.

Barbieri

Levy

(1999) Sleeping with the enemy: The impact of war on trade. Journal of Peace Research 36(4): 463–479.

Barbieri

Keshk

Pollins

(2009) Trading data: Evaluating our assumptions and coding rules. Conflict Management and Peace Science 26(5): 471–491.

Bas

Schub

(2016) How uncertainty about war outcomes affects war onset. Journal of Conflict Resolution 60(6): 1099–1128.

Bas

Schub

(2017) Peaceful uncertainty: When power shocks do not create commitment problems. International Studies Quarterly 61(4): 850–866.

Boehmer

Jungblut

Stoll

(2011) Tradeoffs in trade data do our assumptions affect our results? Conflict Management and Peace Science 28(2): 145–167.

Brooks

(2007) Producing Security: Multinational Corporations, Globalization, and the Changing Calculus of Conflict. Princeton, NJ: Princeton University Press.

10.

Carrubba

Yuen

Zorn

(2007) In defense of comparative statics: Specifying empirical tests of models of strategic interaction. Political Analysis 15(4): 465–482. 65(3): 809–832.

11.

Crescenzi

(2003a) Economic exit, interdependence, and conflict. Journal of Politics 65(3): 809–832.

12.

Crescenzi

(2003b) Interdependence and conflict: When does symmetry matter? Conflict Management and Peace Science 20(1): 73–92.

13.

Crescenzi

(2007) Reputation and interstate conflict. American Journal of Political Science 51(2): 382–396.

14.

Dafoe

Kelsey

(2014) Observing the capitalist peace examining market-mediated signaling and other mechanisms. Journal of Peace Research 51(5): 619–633.

15.

Dorussen

Ward

(2010) Trade networks and the kantian peace. Journal of Peace Research 47(1): 29–42.

16.

Fearon

(1994) Domestic political audiences and the escalation of international disputes. American Political Science Review 88(3): 577–592.

17.

Fearon

(1995) Rationalist explanations for war. International Organization 49(3): 379–414.

18.

Fearon

(1997) Signaling foreign policy interests tying hands versus sinking costs. Journal of Conflict Resolution 41(1): 68–90.

19.

Fey

Ramsay

(2011) Uncertainty and incentives in crisis bargaining: Game-free analysis of international conflict. American Journal of Political Science 55(1): 149–169.

20.

Fforde

(1992) The Bank of England and Public Policy, 1941–1958. Cambridge: Cambridge University Press.

21.

Gartzke

(2007) The capitalist peace. American Journal of Political Science 51(1): 166–191.

22.

Gartzke

(2003) War, peace, and the invisible hand: Positive political externalities of economic globalization. International Studies Quarterly 47(4): 561–586.

23.

Gartzke

Westerwinter

(2016) The complex structure of commercial peace contrasting trade interdependence, asymmetry, and multipolarity. Journal of Peace Research 53(3): 325–343.

24.

Gartzke

Zhang

(2015) Trade and war. In: The Oxford Handbook of the Political Economy of International Trade. Oxford: Oxford University Press, pp. 419–438.

25.

Gartzke

Boehmer

(2001) Investing in the peace: Economic interdependence and international conflict. International Organization 55(2): 391–438.

26.

Gibler

(2009) International Military Alliances, 1648–2008. Washington, DC: CQ Press.

27.

Gibler

Miller

Little

(2016) An analysis of the militarized interstate dispute (MID) dataset, 1816–2001. International Studies Quarterly 60(4): 719–730.

28.

Glick

Taylor

(2010) Collateral damage: Trade disruption and the economic impact of war. The Review of Economics and Statistics 92(1): 102–127.

29.

Gowa

(1994) Allies, Adversaries, and International Trade. Princeton, NJ: Princeton University Press.

30.

Hegre

(2009) Trade dependence or size dependence? The gravity model of trade and the liberal peace. Conflict Management and Peace Science 26(1): 26–45.

31.

Hegre

Oneal

Russett

(2010) Trade does promote peace: New simultaneous estimates of the reciprocal effects of trade and conflict. Journal of Peace Research 47(6): 763–774.

32.

Hirschman

(1980) National Power and the Structure of Foreign Trade, Vol. 105. Berkeley, CA: University of California Press.

33.

Keohane

Nye

(1977) Power and Interdependence: World Politics in Transition. Boston, MA: Little, Brown.

34.

Keshk

Reuveny

Pollins

(2010) Trade and conflict: Proximity, country size, and measures. Conflict Management and Peace Science 27(1): 3–27.

35.

Kinne

(2012) Multilateral trade and militarized conflict: Centrality, openness, and asymmetry in the global trade network. Journal of Politics 74(1): 308–322.

36.

Kinne

(2014) Does third-party trade reduce conflict? Credible signaling versus opportunity costs. Conflict Management and Peace Science 31(1): 28–48.

37.

Kunz

(1991) The Economic Diplomacy of the Suez Crisis. Chapel Hill, NC: University of North Carolina Press.

38.

Long

(2008) Bilateral trade in the shadow of armed conflict. International Studies Quarterly 52(1): 81–101.

39.

Mansfield

Bronson

(1997) Alliances, preferential trading arrangements, and international trade. American Political Science Review 91(1): 94–107.

40.

Mansfield

Milner

Rosendorff

(2000) Free to trade: Democracies, autocracies, and international trade. American Political Science Review 94(2): 305–321.

41.

Marshall

Jaggers

Gurr

(2002) Polity IV Project. Center for International Development and Conflict Management at the University of Maryland, College Park, MD.

42.

McCourt

(2014) Britain and World Power since 1945: Constructing a Nation’s Role in International Politics. Ann Arbor, MI: University of Michigan Press.

43.

Morrow

(1999) How could trade affect conflict? Journal of Peace Research 36(4): 481–489.

44.

Morrow

(2003) Assessing the Role of Trade as a Source of Costly Signals. Economic Interdependence and International Conflict: New Perspectives on an Enduring Debate. Ann Arbor, MI: University of Michigan Press.

45.

Morrow

Siverson

Tabares

(1998) The political determinants of international trade: The major powers, 1907–1990. American Political Science Review 92(3): 649–661.

46.

Mousseau

(2013) The democratic peace unraveled: It’s the economy. International Studies Quarterly 57(1): 186–197.

47.

Oneal

Russett

(1997) The classical liberals were right: Democracy, interdependence, and conflict, 1950–1985. International Studies Quarterly 41(2): 267–294.

48.

Oneal

Russett

(1999) Assessing the liberal peace with alternative specifications: Trade still reduces conflict. Journal of Peace Research 36(4): 423–442.

49.

Opsahl

Agneessens

Skvoretz

(2010) Node centrality in weighted networks: Generalizing degree and shortest paths. Social Networks 32(3): 245–251.

50.

Peterson

(2011) Third-party trade, political similarity, and dyadic conflict. Journal of Peace Research 48(2): 185–200.

51.

Peterson

(2013) Sending a message: The reputation effect of US sanction threat behavior. International Studies Quarterly 57(4): 672–682.

52.

Peterson

(2014) Dyadic trade, exit costs, and conflict. Journal of Conflict Resolution 58(4): 564–591.

53.

Peterson

(2018) Reconsidering economic leverage and vulnerability: Trade ties, sanction threats, and the success of economic coercion. Conflict Management and Peace Science (in press).

54.

Polachek

(1980) Conflict and trade. Journal of Conflict Resolution 24(1): 55–78.

55.

Polachek

Xiang

(2010) How opportunity costs decrease the probability of war in an incomplete information game. International Organization 64(1): 133–144.

56.

Powell

(1990) Nuclear Deterrence Theory: The Search for Credibility. Cambridge University Press.

57.

Powell

(2006) War as a commitment problem. International Organization 60(1): 169–203.

58.

Ramsay

(2008) Settling it on the field battlefield events and war termination. Journal of Conflict Resolution 52(6): 850–879.

59.

Reed

(2003a) Information and economic interdependence. Journal of Conflict Resolution 47(1): 54–71.

60.

Reed

(2003b) Information, power, and war. American Political Science Review 97(4): 633–641.

61.

Rosecrance

(1986) The Rise of the Trading State: Commerce and Conquest in the Modern World. New York: Basic Books.

62.

Sartori

(2002) The might of the pen: A reputational theory of communication in international disputes. International Organization 56(1): 121–149.

63.

Sartori

(2003) An estimator for some binary-outcome selection models without exclusion restrictions. Political Analysis 11(2): 111–138.

64.

Schultz

(1998) Domestic opposition and signaling in international crises. American Political Science Review 92(4): 829–844.

65.

Schultz

(2001a) Democracy and Coercive Diplomacy, Vol. 76. Cambridge: Cambridge University Press.

66.

Schultz

(2001b) Looking for audience costs. Journal of Conflict Resolution 45(1): 32–60.

67.

Signorino

(1999) Strategic interaction and the statistical analysis of international conflict. American Political Science Review 93(2): 279–297.

68.

Signorino

(2007) On formal theory and statistical methods: A response to Carrubba, Yuen, and Zorn. Political Analysis 15(4): 483–501.

69.

Singer

Bremer

Stuckey

(1972) Capability distribution, uncertainty, and major power war, 1820–1965. Peace, War, and Numbers 19: 48.

70.

Slantchev

(2004) How initiators end their wars: The duration of warfare and the terms of peace. American Journal of Political Science 48(4): 813–829.

71.

Slantchev

(2005) Military coercion in interstate crises. American Political Science Review 99(4): 533–547.

72.

Slantchev

(2011) Military Threats: The Costs of Coercion and the Price of Peace. Cambridge: Cambridge University Press.

73.

Smith

(1999) Testing theories of strategic choice: The example of crisis escalation. American Journal of Political Science 43(4): 1254–1283.

74.

Steed

(2016) British Strategy and Intelligence in the Suez Crisis. Berlin: Springer.

75.

Stein

(1993) Governments, Economic Interdependence, and International Cooperation. Behavior, Society, and International Conflict. New York: Oxford University Press.

76.

Stinnett

Tir

Diehl

Schafer

Gochman

(2002) The Correlates of War (COW) Project Direct Contiguity Data, version 3.0. Conflict Management and Peace Science 19(2): 59–67.

77.

Turner

(2014) Macmillan. New York: Routledge.

78.

Wagner

(1988) Economic interdependence, bargaining power, and political influence. International Organization 42(3): 461–483.

79.

Ward

Hoff

(2007) Persistent patterns of international commerce. Journal of Peace Research 44(2): 157–175.

80.

Whang

(2010) Empirical implications of signaling models: Estimation of belief updating in international crisis bargaining. Political Analysis 18(3): 381–402.

81.

Whang

McLean

Kuberski

(2013) Coercion, information, and the success of sanction threats. American Journal of Political Science 57(1): 65–81.

82.

Wolford

(2014) Power, preferences, and balancing: The durability of coalitions and the expansion of conflict. International Studies Quarterly 58(1): 146–157.

83.

Wolford

Reiter

Carrubba

(2011) Information, commitment, and war. Journal of Conflict Resolution 55(4): 556–579.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

6.97 MB

0.00 MB

0.68 MB